hpricot 0.4-mswin32 → 0.5-mswin32

data/ext/hpricot_scan/hpricot_scan.rl

@@ -8,14 +8,21 @@
  */
 #include <ruby.h>
 
+#define NO_WAY_SERIOUSLY "*** This should not happen, please send a bug report with the HTML you're parsing to why@whytheluckystiff.net.  So sorry!"
+
 static VALUE sym_xmldecl, sym_doctype, sym_procins, sym_stag, sym_etag, sym_emptytag, sym_comment,
       sym_cdata, sym_text;
+static VALUE rb_eHpricotParseError;
 static ID s_read, s_to_str;
 
 #define ELE(N) \
-  if (tokend > tokstart) { \
-    ele_open = 0; \
-    rb_yield_tokens(sym_##N, tag, attr, tokstart == 0 ? Qnil : rb_str_new(tokstart, tokend-tokstart), taint); \
+  if (tokend > tokstart || text == 1) { \
+    VALUE raw_string = Qnil; \
+    ele_open = 0; text = 0; \
+    if (tokstart != 0 && sym_##N != sym_cdata && sym_##N != sym_text && sym_##N != sym_procins && sym_##N != sym_comment) { \
+      raw_string = rb_str_new(tokstart, tokend-tokstart); \
+    } \
+    rb_yield_tokens(sym_##N, tag, attr, raw_string, taint); \
   }
 
 #define SET(N, E) \
@@ -34,6 +41,24 @@ static ID s_read, s_to_str;
       rb_hash_aset(attr, K, V); \
     }
 
+#define TEXT_PASS() \
+    if (text == 0) \
+    { \
+      if (ele_open == 1) { \
+        ele_open = 0; \
+        if (tokstart > 0) { \
+          mark_tag = tokstart; \
+        } \
+      } else { \
+        mark_tag = p; \
+      } \
+      attr = Qnil; \
+      tag = Qnil; \
+      text = 1; \
+    }
+
+#define EBLK(N, T) CAT(tag, p - T + 1); ELE(N);
+
 %%{
   machine hpricot_scan;
 
@@ -55,6 +80,10 @@ static ID s_read, s_to_str;
   action tag { SET(tag, p); }
   action tagc { SET(tag, p-1); }
   action aval { SET(aval, p); }
+  action aunq { 
+    if (*(p-1) == '"' || *(p-1) == '\'') { SET(aval, p-1); }
+    else { SET(aval, p); }
+  }
   action akey { SET(akey, p); }
   action xmlver { SET(aval, p); ATTR(rb_str_new2("version"), aval); }
   action xmlenc { SET(aval, p); ATTR(rb_str_new2("encoding"), aval); }
@@ -79,7 +108,7 @@ static ID s_read, s_to_str;
   #
   newline = '\n' @{curline += 1;} ;
 # qtext = '"' ( '\"' | [^\n"] )* '"' | "'" ( "\\'" | [^\n'] )* "'" ; 
-  NameChar = [\-A-Za-z0-9._:] ;
+  NameChar = [\-A-Za-z0-9._:?] ;
   Name = [A-Za-z_:] NameChar* ;
   StartComment = "<!--" ;
   EndComment = "-->" ;
@@ -87,14 +116,14 @@ static ID s_read, s_to_str;
   EndCdata = "]]>" ;
 
   NameCap = Name >_tag %tag;
-  NameAttr = Name >_akey %akey ;
+  NameAttr = NameChar+ >_akey %akey ;
   Q1Attr = [^']* >_aval %aval ;
   Q2Attr = [^"]* >_aval %aval ;
-  UnqAttr = [^ \t\n<>"'] >_aval [^ \t\n<>]* %aval ;
+  UnqAttr = ( space >_aval | [^ \t\n<>"'] >_aval [^ \t\n<>]* %aunq ) ;
   Nmtoken = NameChar+ >_akey %akey ;
 
   Attr =  NameAttr space* "=" space* ('"' Q2Attr '"' | "'" Q1Attr "'" | UnqAttr space+ ) space* ;
-  AttrEnd = ( NameAttr space* "=" space* UnqAttr | Nmtoken >new_attr %save_attr ) ;
+  AttrEnd = ( NameAttr space* "=" space* UnqAttr? | Nmtoken >new_attr %save_attr ) ;
   AttrSet = ( Attr >new_attr %save_attr | Nmtoken >new_attr space+ %save_attr ) ;
   StartTag = "<" NameCap space+ AttrSet* (AttrEnd >new_attr %save_attr)? ">" | "<" NameCap ">";
   EmptyTag = "<" NameCap space+ AttrSet* (AttrEnd >new_attr %save_attr)? "/>" | "<" NameCap "/>" ;
@@ -113,14 +142,23 @@ static ID s_read, s_to_str;
     "'" [\t a-zA-Z0-9\-'()+,./:=?;!*\#@$_%]* >_aval %pubid "'" ;
   ExternalID = ( "SYSTEM" | "PUBLIC" space+ PubidLiteral ) (space+ SystemLiteral)? ;
   DocType = "<!DOCTYPE" space+ NameCap (space+ ExternalID)? space* ("[" [^\]]* "]" space*)? ">" ;
-  StartXmlProcIns = "<?" Name space+ ;
+  StartXmlProcIns = "<?" Name >{ TEXT_PASS(); } space+ ;
   EndXmlProcIns = "?>" ;
 
-  html_comment := (any | newline )* >_tag :>> EndComment >tagc @{ ELE(comment); fgoto main; };
+  html_comment := |*
+    EndComment @{ EBLK(comment, 3); fgoto main; };
+    any | newline { TEXT_PASS(); };
+  *|;
 
-  html_cdata := (any | newline )* >_tag :>> EndCdata >tagc @{ ELE(cdata); fgoto main; };
+  html_cdata := |*
+    EndCdata @{ EBLK(cdata, 3); fgoto main; };
+    any | newline { TEXT_PASS(); };
+  *|;
 
-  html_procins := (any | newline )* >_tag :>> EndXmlProcIns >tagc @{ ELE(procins); fgoto main; };
+  html_procins := |*
+    EndXmlProcIns @{ EBLK(procins, 2); fgoto main; };
+    any | newline { TEXT_PASS(); };
+  *|;
 
   main := |*
     XmlDecl >newEle { ELE(xmldecl); };
@@ -131,23 +169,7 @@ static ID s_read, s_to_str;
     EmptyTag >newEle { ELE(emptytag); };
     StartComment >newEle { fgoto html_comment; };
     StartCdata >newEle { fgoto html_cdata; };
-
-    any | newline {
-      if (text == 0)
-      {
-        if (ele_open == 1) {
-          ele_open = 0;
-          if (tokstart > 0) {
-            mark_tag = tokstart;
-          }
-        } else {
-          mark_tag = p;
-        }
-        attr = Qnil;
-        tag = Qnil;
-        text = 1;
-      }
-    };
+    any | newline { TEXT_PASS(); };
   *|;
 }%%
 
@@ -173,13 +195,12 @@ void rb_yield_tokens(VALUE sym, VALUE tag, VALUE attr, VALUE raw, int taint)
 
 VALUE hpricot_scan(VALUE self, VALUE port)
 {
-  static char buf[BUFSIZE];
   int cs, act, have = 0, nread = 0, curline = 1, text = 0;
-  char *tokstart = 0, *tokend = 0;
+  char *tokstart = 0, *tokend = 0, *buf = NULL;
 
-  VALUE attr = Qnil, tag = Qnil, akey = Qnil, aval = Qnil;
+  VALUE attr = Qnil, tag = Qnil, akey = Qnil, aval = Qnil, bufsize = Qnil;
   char *mark_tag = 0, *mark_akey = 0, *mark_aval = 0;
-  int done = 0, ele_open = 0;
+  int done = 0, ele_open = 0, buffer_size = 0;
 
   int taint = OBJ_TAINTED( port );
   if ( !rb_respond_to( port, s_read ) )
@@ -195,18 +216,27 @@ VALUE hpricot_scan(VALUE self, VALUE port)
     }
   }
 
+  buffer_size = BUFSIZE;
+  if (rb_ivar_defined(self, rb_intern("@buffer_size")) == Qtrue) {
+    bufsize = rb_ivar_get(self, rb_intern("@buffer_size"));
+    if (!NIL_P(bufsize)) {
+      buffer_size = NUM2INT(bufsize);
+    }
+  }
+  buf = ALLOC_N(char, buffer_size);
+
   %% write init;
   
   while ( !done ) {
     VALUE str;
     char *p = buf + have, *pe;
-    int len, space = BUFSIZE - have;
+    int len, space = buffer_size - have;
 
     if ( space == 0 ) {
       /* We've used up the entire buffer storing an already-parsed token
-       * prefix that must be preserved. */
-      fprintf(stderr, "OUT OF BUFFER SPACE\n" );
-      exit(1);
+       * prefix that must be preserved.  Likely caused by super-long attributes.
+       * See ticket #13. */
+      rb_raise(rb_eHpricotParseError, "ran out of buffer space on element <%s>, starting on line %d.", RSTRING(tag)->ptr, curline);
     }
 
     if ( rb_respond_to( port, s_read ) )
@@ -233,8 +263,15 @@ VALUE hpricot_scan(VALUE self, VALUE port)
     %% write exec;
     
     if ( cs == hpricot_scan_error ) {
-      fprintf(stderr, "PARSE ERROR\n" );
-      break;
+      free(buf);
+      if ( !NIL_P(tag) )
+      {
+        rb_raise(rb_eHpricotParseError, "parse error on element <%s>, starting on line %d.\n" NO_WAY_SERIOUSLY, RSTRING(tag)->ptr, curline);
+      }
+      else
+      {
+        rb_raise(rb_eHpricotParseError, "parse error on line %d.\n" NO_WAY_SERIOUSLY, curline);
+      }
     }
     
     if ( done && ele_open )
@@ -279,12 +316,15 @@ VALUE hpricot_scan(VALUE self, VALUE port)
       tokstart = buf;
     }
   }
+  free(buf);
 }
 
 void Init_hpricot_scan()
 {
   VALUE mHpricot = rb_define_module("Hpricot");
+  rb_define_attr(rb_singleton_class(mHpricot), "buffer_size", 1, 1);
   rb_define_singleton_method(mHpricot, "scan", hpricot_scan, 1);
+  rb_eHpricotParseError = rb_define_class_under(mHpricot, "ParseError", rb_eException);
 
   s_read = rb_intern("read");
   s_to_str = rb_intern("to_str");

data/lib/hpricot.rb

@@ -1,3 +1,22 @@
+# == About hpricot.rb
+#
+# All of Hpricot's various part are loaded when you use <tt>require 'hpricot'</tt>.
+#
+# * hpricot_scan: the scanner (a C extension for Ruby) which turns an HTML stream into tokens.
+# * hpricot/parse.rb: uses the scanner to sort through tokens and give you back a complete document object.
+# * hpricot/tag.rb: sets up objects for the various types of elements in an HTML document.
+# * hpricot/modules.rb: categorizes the various elements using mixins.
+# * hpricot/traverse.rb: methods for searching documents.
+# * hpricot/elements.rb: methods for dealing with a group of elements as an Hpricot::Elements list.
+# * hpricot/inspect.rb: methods for displaying documents in a readable form.
+
+# If available, Nikolai's UTF-8 library will ease use of utf-8 documents.
+# See http://git.bitwi.se/ruby-character-encodings.git/.
+begin
+  require 'encoding/character/utf-8'
+rescue LoadError
+end
+
 require 'hpricot_scan'
 require 'hpricot/tag'
 require 'hpricot/modules'

data/lib/hpricot/elements.rb

@@ -1,66 +1,163 @@
 module Hpricot
+# Once you've matched a list of elements, you will often need to handle them as
+# a group.  Or you may want to perform the same action on each of them.
+# Hpricot::Elements is an extension of Ruby's array class, with some methods
+# added for altering elements contained in the array.
+#
+# If you need to create an element array from regular elements:
+#
+#   Hpricot::Elements[ele1, ele2, ele3]
+#
+# Assuming that ele1, ele2 and ele3 contain element objects (Hpricot::Elem,
+# Hpricot::Doc, etc.)  
+#
+# == Continuing Searches
+#
+# Usually the Hpricot::Elements you're working on comes from a search you've
+# done.  Well, you can continue searching the list by using the same <tt>at</tt>
+# and <tt>search</tt> methods you can use on plain elements.
+#
+#   elements = doc.search("/div/p")
+#   elements = elements.search("/a[@href='http://hoodwink.d/']")
+#   elements = elements.at("img")
+#
+# == Altering Elements
+#
+# When you're altering elements in the list, your changes will be reflected in
+# the document you started searching from.
+#
+#   doc = Hpricot("That's my <b>spoon</b>, Tyler.")
+#   doc.at("b").swap("<i>fork</i>")
+#   doc.to_html
+#     #=> "That's my <i>fork</i>, Tyler." 
+#
+# == Getting More Detailed
+#
+# If you can't find a method here that does what you need, you may need to
+# loop through the elements and find a method in Hpricot::Container::Trav
+# which can do what you need.
+#
+# For example, you may want to search for all the H3 header tags in a document
+# and grab all the tags underneath the header, but not inside the header.
+# A good method for this is <tt>next_sibling</tt>:
+#
+#   doc.search("h3").each do |h3|
+#     while ele = h3.next_sibling
+#       ary << ele   # stuff away all the elements under the h3
+#     end
+#   end
+#
+# Most of the useful element methods are in the mixins Hpricot::Traverse
+# and Hpricot::Container::Trav.
   class Elements < Array
+    # Searches this list for any elements (or children of these elements) matching
+    # the CSS or XPath expression +expr+.  Root is assumed to be the element scanned.
+    #
+    # See Hpricot::Container::Trav.search for more.
     def search(*expr,&blk)
-      map { |x| x.search(*expr,&blk) }.flatten.uniq
+      Elements[*map { |x| x.search(*expr,&blk) }.flatten.uniq]
     end
     alias_method :/, :search
 
+    # Searches this list for the first element (or child of these elements) matching
+    # the CSS or XPath expression +expr+.  Root is assumed to be the element scanned.
+    #
+    # See Hpricot::Container::Trav.at for more.
     def at(expr, &blk)
       search(expr, &blk).first
     end
     alias_method :%, :at
 
+    # Convert this group of elements into a complete HTML fragment, returned as a
+    # string.
     def to_html
       map { |x| x.output("") }.join
     end
     alias_method :to_s, :to_html
 
-    def inner_html(*str)
-      if str.empty?
+    # Returns an HTML fragment built of the contents of each element in this list.
+    #
+    # If a HTML +string+ is supplied, this method acts like inner_html=.
+    def inner_html(*string)
+      if string.empty?
         map { |x| x.inner_html }.join
       else
-        x = self.inner_html = str.pop || x
+        x = self.inner_html = string.pop || x
       end
     end
-    alias_method :text, :inner_html
     alias_method :html, :inner_html
     alias_method :innerHTML, :inner_html
 
-    def inner_html=(str)
-      each { |x| x.inner_html = str }
+    # Replaces the contents of each element in this list.  Supply an HTML +string+,
+    # which is loaded into Hpricot objects and inserted into every element in this
+    # list.
+    def inner_html=(string)
+      each { |x| x.inner_html = string }
     end
     alias_method :html=, :inner_html=
     alias_method :innerHTML=, :inner_html=
 
-    def filter(expr)
-        nodes, = Elements.filter(self, expr)
-        nodes
+    # Returns an string containing the text contents of each element in this list.
+    # All HTML tags are removed.
+    def inner_text
+      map { |x| x.inner_text }.join
     end
+    alias_method :text, :inner_text
 
+    # Remove all elements in this list from the document which contains them.
+    #
+    #   doc = Hpricot("<html>Remove this: <b>here</b></html>")
+    #   doc.search("b").remove
+    #   doc.to_html
+    #     => "<html>Remove this: </html>"
+    #
     def remove
       each { |x| x.parent.children.delete(x) }
     end
 
+    # Empty the elements in this list, by removing their insides.
+    #
+    #   doc = Hpricot("<p> We have <i>so much</i> to say.</p>")
+    #   doc.search("i").empty
+    #   doc.to_html
+    #     => "<p> We have <i></i> to say.</p>"
+    #
     def empty
       each { |x| x.inner_html = nil }
     end
 
+    # Add to the end of the contents inside each element in this list.
+    # Pass in an HTML +str+, which is turned into Hpricot elements.
     def append(str)
       each { |x| x.inner_html += str }
     end
 
+    # Add to the start of the contents inside each element in this list.
+    # Pass in an HTML +str+, which is turned into Hpricot elements.
     def prepend(str)
       each { |x| x.inner_html = str + x.inner_html }
     end
-
+ 
+    # Add some HTML just previous to each element in this list.
+    # Pass in an HTML +str+, which is turned into Hpricot elements.
     def before(str)
       each { |x| x.parent.insert_before Hpricot.make(str), x }
     end
 
+    # Just after each element in this list, add some HTML.
+    # Pass in an HTML +str+, which is turned into Hpricot elements.
     def after(str)
       each { |x| x.parent.insert_after Hpricot.make(str), x }
     end
 
+    # Wraps each element in the list inside the element created by HTML +str+. 
+    # If more than one element is found in the string, Hpricot locates the
+    # deepest spot inside the first element.
+    #
+    #  doc.search("a[@href]").
+    #      wrap(%{<div class="link"><div class="link_inner"></div></div>})
+    #
+    # This code wraps every link on the page inside a +div.link+ and a +div.link_inner+ nest.
     def wrap(str)
       each do |x|
         wrap = Hpricot.make(str)
@@ -74,15 +171,15 @@ module Hpricot
       end
     end
 
-    def not(expr)
-        if expr.is_a? Container::Trav
-            nodes = self - [expr]
-        else
-            nodes, = Elements.filter(self, expr, false)
-        end
-        nodes
-    end
-
+    # Sets an attribute for all elements in this list.  You may use
+    # a simple pair (<em>attribute name</em>, <em>attribute value</em>):
+    #
+    #   doc.search('p').set(:class, 'outline')
+    #
+    # Or, use a hash of pairs:
+    #
+    #   doc.search('div#sidebar').set(:class => 'outline', :id => 'topbar')
+    #
     def set(k, v = nil)
       case k
       when Hash
@@ -96,9 +193,9 @@ module Hpricot
       end
     end
 
-    ATTR_RE = %r!\[ *(@)([a-zA-Z0-9\(\)_-]+) *([~\!\|\*$\^=]*) *'?"?([^'"]*)'?"? *\]!i
+    ATTR_RE = %r!\[ *(?:(@)([\w\(\)-]+)|([\w\(\)-]+\(\))) *([~\!\|\*$\^=]*) *'?"?([^'"]*)'?"? *\]!i
     BRACK_RE = %r!(\[) *([^\]]*) *\]!i
-    FUNC_RE = %r!(:)([a-zA-Z0-9\*_-]*)\( *[\"']?([^ \)'\"]*)['\"]? *\)!
+    FUNC_RE = %r!(:)?([a-zA-Z0-9\*_-]*)\( *[\"']?([^ \)]*?)['\"]? *\)!
     CATCH_RE = %r!([:\.#]*)([a-zA-Z0-9\*_-]+)!
 
     def self.filter(nodes, expr, truth = true)
@@ -112,16 +209,20 @@ module Hpricot
                 m[0] = "@#{m.slice!(2,1)}"
             end
 
+            if m[0] == '[' && m[1] =~ /^\d+$/
+                m = [":", "nth", m[1].to_i-1]
+            end
+
             if m[0] == ":" && m[1] == "not"
                 nodes, = Elements.filter(nodes, m[2], false)
             else
-                meth = "filter[#{m[0]}]"
-                if Container::Trav.method_defined? meth
-                    args = m[1..-1]
+                meth = "filter[#{m[0]}#{m[1]}]"
+                if Traverse.method_defined? meth
+                    args = m[2..-1]
                 else
-                    meth = "filter[#{m[0]}#{m[1]}]"
-                    if Container::Trav.method_defined? meth
-                        args = m[2..-1]
+                    meth = "filter[#{m[0]}]"
+                    if Traverse.method_defined? meth
+                        args = m[1..-1]
                     end
                 end
                 i = -1
@@ -134,7 +235,19 @@ module Hpricot
         [nodes, expr]
     end
 
-    def inspect; "#<#{self.class}#{super}>" end
+    def filter(expr)
+        nodes, = Elements.filter(self, expr)
+        nodes
+    end
+
+    def not(expr)
+        if expr.is_a? Traverse
+            nodes = self - [expr]
+        else
+            nodes, = Elements.filter(self, expr, false)
+        end
+        nodes
+    end
 
     private
     def copy_node(node, l)
@@ -145,50 +258,51 @@ module Hpricot
 
   end
 
-  module Container::Trav
+  module Traverse
     def self.filter(tok, &blk)
       define_method("filter[#{tok.is_a?(String) ? tok : tok.inspect}]", &blk)
     end
 
     filter '' do |name,i|
-      name == '*' || self.name.downcase == name.downcase
+      name == '*' || (self.respond_to?(:name) && self.name.downcase == name.downcase)
     end
 
     filter '#' do |id,i|
-      get_attribute('id').to_s == id
+      self.elem? and get_attribute('id').to_s == id
     end
 
     filter '.' do |name,i|
-      classes.include? name
+      self.elem? and classes.include? name
     end
 
     filter :lt do |num,i|
-      parent.containers.index(self) < num.to_i
+      self.position < num.to_i
     end
 
     filter :gt do |num,i|
-      parent.containers.index(self) > num.to_i
+      self.position > num.to_i
     end
 
-    nth = proc { |num,i| parent.containers.index(self) == num.to_i }
+    nth = proc { |num,i| self.position == num.to_i }
+    nth_first = proc { |*a| self.position == 0 }
+    nth_last = proc { |*a| self == parent.children_of_type(self.name).last }
 
     filter :nth, &nth
     filter :eq, &nth
+    filter ":nth-of-type", &nth
 
-    filter :first do |num,i|
-      parent.containers.index(self) == 0
-    end
+    filter :first, &nth_first
+    filter ":first-of-type", &nth_first
 
-    filter :last do |i|
-      self == parent.containers.last
-    end
+    filter :last, &nth_last
+    filter ":last-of-type", &nth_last
 
     filter :even do |num,i|
-      parent.containers.index(self) % 2 == 0
+      self.position % 2 == 0
     end
 
     filter :odd do |num,i|
-      parent.containers.index(self) % 2 == 1
+      self.position % 2 == 1
     end
 
     filter ':first-child' do |i|
@@ -204,32 +318,19 @@ module Hpricot
     end
 
     filter ":last-child" do |i|
-      self == parent.containers.first
+      self == parent.containers.last
     end
     
     filter ":nth-last-child" do |arg,i|
       self == parent.containers[-1-arg.to_i]
     end
 
-    filter ":first-of-type" do |i|
-      self == parent.containers.detect { |x| x.name == arg }
-    end
-
-    filter ":nth-of-type" do |arg,i|
-      self == parent.containers.find_all { |x| x.name == arg }[arg.to_i]
-    end
-
-    filter ":last-of-type" do |i|
-      self == parent.containers.find_all { |x| x.name == self.name }.last
-    end
-
-    filter :"nth-last-of-type" do |arg,i|
-      self == parent.containers.find_all { |x| x.name == arg }[-1-arg.to_i]
+    filter ":nth-last-of-type" do |arg,i|
+      self == parent.children_of_type(self.name)[-1-arg.to_i]
     end
 
     filter ":only-of-type" do |arg,i|
-      of_type = parent.containers.find_all { |x| x.name == arg }
-      of_type.length == 1
+      parent.children_of_type(self.name).length == 1
     end
 
     filter ":only-child" do |arg,i|
@@ -237,55 +338,61 @@ module Hpricot
     end
 
     filter :parent do
-      childNodes.length > 0
+      containers.length > 0
     end
 
     filter :empty do
-      childNodes.length == 0
+      containers.length == 0
     end
 
     filter :root do
       self.is_a? Hpricot::Doc
     end
     
-    filter :contains do |arg,|
-      html.include? arg
-    end
-
-    filter '@=' do |attr,val,i|
-      get_attribute(attr).to_s == val
-    end
-
-    filter '@!=' do |attr,val,i|
-      get_attribute(attr).to_s != val
-    end
-
-    filter '@~=' do |attr,val,i|
-      get_attribute(attr).to_s.split(/\s+/).include? val
+    filter 'text' do
+      self.text?
     end
 
-    filter '@|=' do |attr,val,i|
-      get_attribute(attr).to_s =~ /^#{Regexp::quote val}(-|$)/
+    filter 'comment' do
+      self.comment?
     end
 
-    filter '@^=' do |attr,val,i|
-      get_attribute(attr).to_s.index(val) == 0
+    filter :contains do |arg,|
+      html.include? arg
     end
 
-    filter '@$=' do |attr,val,i|
-      get_attribute(attr).to_s =~ /#{Regexp::quote val}$/
+    pred_procs =
+      {'text()' => proc { |ele, *_| ele.inner_text.strip },
+       '@'      => proc { |ele, attr, *_| ele.get_attribute(attr).to_s if ele.elem? }}
+    
+    oper_procs =
+      {'='      => proc { |a,b| a == b },
+       '!='     => proc { |a,b| a != b },
+       '~='     => proc { |a,b| a.split(/\s+/).include?(b) },
+       '|='     => proc { |a,b| a =~ /^#{Regexp::quote b}(-|$)/ },
+       '^='     => proc { |a,b| a.index(b) == 0 },
+       '$='     => proc { |a,b| a =~ /#{Regexp::quote b}$/ },
+       '*='     => proc { |a,b| idx = a.index(b) }}
+
+    pred_procs.each do |pred_n, pred_f|
+      oper_procs.each do |oper_n, oper_f|
+        filter "#{pred_n}#{oper_n}" do |*a|
+          qual = pred_f[self, *a]
+          oper_f[qual, a[-2]] if qual
+        end
+      end
     end
 
-    filter '@*=' do |attr,val,i|
-      get_attribute(attr).to_s.index(val) >= 0
+    filter 'text()' do |val,i|
+      !self.inner_text.strip.empty?
     end
 
     filter '@' do |attr,val,i|
-      has_attribute? attr
+      self.elem? and has_attribute? attr
     end
 
     filter '[' do |val,i|
-      search(val).length > 0
+      self.elem? and search(val).length > 0
     end
 
   end